Skip to content
Xavier WANG

MIND · MODELS · MOTION

AI FOR DISASTER INTELLIGENCE

XAVIER WANG

Understanding Hazard Dynamics with World Models

I study how world models represent and anticipate change in complex hazard environments, supporting multi-hazard forecasting, risk warning, and intelligent decision-making in real settings.

Explore ResearchView ProjectsGet in Touch

Think deeply, model change, and understand a world in motion.

ENVIRONMENT OBSERVATION·RAINFALL INPUT

Research

Current Research Focus

How can intelligent systems understand and anticipate a physical world that is constantly changing and inherently uncertain?

Current Focus

Disaster Intelligence

Studying landslides, debris flows, and floods through remote sensing, geospatial information, spatiotemporal data, and machine learning for risk sensing, early warning, and decision support.

Read more
Research Direction

World Models

Exploring models that represent environmental states, learn evolving dynamics, and simulate possible futures for complex hazard processes.

Read more
Methods and Exploration

Generative AI

Exploring generative models for scientific data modeling, limited-data augmentation, cross-modal understanding, and dynamic process simulation.

Read more

Research Journey

From Computer Science to Disaster Intelligence

  1. 2020 — 2024

    Computer Science

    The Hong Kong Polytechnic University

    A foundation in algorithms, systems, and software engineering.

  2. 2021 — 2023

    Software and Data Systems

    Engineering practice

    Building backends, databases, and data pipelines that work in practice.

  3. 2023 — 2024

    IoT and Data Analytics

    LoRaWAN project

    From device-level data collection and transmission to modeling and analysis.

  4. 2024 — 2025

    Generative AI and Medical Video

    The Hong Kong Polytechnic University

    Modeling dynamic scientific data — from clinical text to angiography videos (JAMA Ophthalmology, 2025).

  5. 2024 — Present

    Disaster Intelligence

    Chengdu

    Multi-hazard early warning and rainfall modeling for real-world decisions — toward world models for complex environmental systems.

02Selected Work

Selected Systems and Projects

From research questions to models, networks, and systems—with evidence and current access status stated transparently.

RESEARCH & ENGINEERING · 2021—PRESENT

Applied Research

2024—Present · Doctoral Research

Current Focus

Flagship Outcome · RainGod-100

AI for Multi-hazard Early Warning and Rainfall Modeling

Fuses multisource observations with AI to forecast rainfall up to two hours ahead for multi-hazard early warning.

  • Precipitation Nowcasting
  • Observation Fusion
  • Multi-hazard Warning
RainGod-100 ten-minute rainfall-intensity forecast over China on August 11, 2026
Real forecast example · RainGod-100 · Aug 11, 2026

Applied Research

JAMA Ophthalmology · Co-first Author

Fundus Fluorescein Angiography Video Generation

Generates dynamic FFA sequences from clinical text to explore privacy-aware medical data sharing.

TEXT CONDITION · MICROANEURYSMS · LEAKAGE · HEMORRHAGE · NEOVASCULARIZATION

  • Generative AI
  • Medical Video
  • Privacy-aware Sharing
MODEL-GENERATED FFA SEQUENCE · NOT REAL PATIENT FOOTAGE
Concept visual of LoRaWAN sensors, a gateway, and an analytics system in mountainous terrain
CONCEPT VISUAL

Data Systems

LoRaWAN-based IoT Communication and Data Analytics System

Built a LoRaWAN pipeline for device data collection, transmission, management, visualization, and exploratory modeling.

2023—2024 · Design & Development

  • LoRaWAN
  • IoT Communications
  • Data Analytics
Concept visual of blockchain nodes, a digital wallet, and secure interactions
CONCEPT VISUAL

Software Engineering

Blockchain and Digital Wallet System

Built a blockchain and digital wallet system, then verified deployment and on-chain interaction on an Ethereum-compatible network.

Historical note: the associated service, nodes, and chain address are no longer accessible.

  • Blockchain
  • Digital Wallet
  • On-chain Interaction
View all projects

03

Featured Publication

07Contact

Contact & Collaboration

I welcome conversations around disaster intelligence, world models, generative scientific data, and interdisciplinary engineering. A collaboration can begin with a shared question, a real dataset, or an idea worth testing.

Collaboration

Start with a real question

On-site messaging is not open yet. Research and engineering conversations are still welcome; you can check the contact page for availability or see how to request a full CV.

Multi-hazard forecasting & risk warningWorld models for environmental systemsGenerative scientific dataIntelligent systems & engineering
View contact detailsFull CV information

Affiliation

Academic identity

Emblem of the Chinese Academy of SciencesMark of the Institute of Mountain Hazards and EnvironmentPersonal academic site · Affiliation information

Institute of Mountain Hazards and Environment, Chinese Academy of Sciences

Doctoral Researcher

Collaboration areas

Exploring intelligence in complex environments

I welcome conversations with researchers, students, and engineers in adjacent fields, especially around modeling, forecasting, and decision problems grounded in real settings.

  • Multi-hazard forecasting & risk warning
  • World models for environmental systems
  • Generative scientific data
  • Intelligent systems & engineering
Explore research directions